UK retail is one of the fastest-moving AI adoption stories in the country. Marks & Spencer’s Copilot rollout to 35,000 colleagues was the headline number that broke through in 2025. Underneath it sat a much bigger pattern: every major UK retailer is now running AI somewhere in the operating model. Some have shipped agentic forecasting; some are deep into AI-driven merchandising; all are wrestling with the talent question that comes after the licence purchase.
The licence is the easy part. The hard part is the people who can design retail-specific AI workflows, govern them across thousands of stores, and make them stick in a workforce that turns over at retail rates. That’s a different skills problem to the one most retail HR directors have ever commissioned a training programme for. Apprenticeships solve more of it than the sector usually realises.
The five AI-skill populations every UK retailer has
1. Head-office AI leadership, the CIO/COO/CDO layer making the platform decisions.
2. Trading & merchandising, the people running the AI forecasting, allocation and pricing tools.
3. Store & field ops, the area managers and store managers who deploy Copilot/Gemini/Claude at the customer face.
4. Supply chain & logistics, the planners using AI for demand forecasting and load optimisation.
5. Customer service & e-comm, the AI assistant, chat and personalisation function.
What the headline rollouts actually tell us
Marks & Spencer, Microsoft Copilot to 35,000 colleagues
The M&S rollout was the largest Copilot deployment in UK retail in 2025. The interesting bit isn’t the licence count, it’s what happened next. The deployment hit familiar walls: variable adoption depth by function, governance gaps around customer data, and a missing layer of internal capability to actually build the bespoke workflows that turn the licence into measurable productivity. The same pattern shows up across most large UK retail Copilot deployments.
John Lewis, agentic forecasting
JLP’s investments in agentic AI for forecasting and supply-chain optimisation are running ahead of most of the sector. The capability constraint there is in the data-and-AI engineering function, the population the ST1398 Machine Learning Engineer L6 standard is explicitly designed for.
Sainsbury’s and Tesco, supplier and supply-chain AI
Both grocers are running AI deep into supplier negotiations, range optimisation and demand forecasting. Most of the capability gap is in the merchandising and trading population, which is the natural fit for ST1512 with the right unit selection.
Which standards fit which population
The portfolio breaks down cleanly across the five populations above:
AU0009, AU0010, AU0011, for the head-office leadership layer
Three Level 5 units (£750 each), stackable over 6–9 months. AU0009 is the strategic case, AU0010 is governance (essential for retailers handling customer data at scale), AU0011 is applied adoption. Best fit: CIO/COO/CDO direct reports, retail-ops directors, customer-data leads.
ST1512, for the trading, merchandising, and store-ops layer
Level 4 apprenticeship, 15 months, £0 for SMEs or up to £18,000 of levy spend. Best fit: trading managers, merchandisers, area managers being upskilled into AI workflow design, and store-ops generalists with cross-channel remits.
ST1398, for the in-house data & AI engineering team
Level 6, 24 months, up to £26,000 of levy spend. Best fit: the in-house data scientists and ML engineers behind forecasting, personalisation, supply-chain optimisation. JLP, the grocers and the larger fashion players all have this function now.
Retail clients we work with usually start with the same brief: “We’ve got Copilot deployed and we’re seeing variable adoption.” The honest answer is almost always that the licence isn’t the problem. The missing layer is the people who design the workflows the licence enables. ST1512 fills exactly that gap. , Rod Doyle, Director, TESS Group
Worked example: 8,000-colleague UK retailer
Three-tier programme stack, levy-funded
Tier 1: 12 senior leaders on AU0009/10/11 (~£27,000 levy spend).
Tier 2: 8 trading and merchandising managers on ST1512 (~£144,000 levy spend).
Tier 3: 2 in-house AI engineers on ST1398 (~£52,000 levy spend).
Plus: closed-cohort Build AI Agents workshop (£4,500 cash) for the e-comm team.
Total levy spend: ~£223,000 over 18 months. Capability output: firmwide governance, eight production AI workflows in trading and merchandising, in-house engineering capability for forecasting and personalisation.
The three patterns we see most
Pattern 1, Copilot rollout, no internal capability. Licences live, adoption variable, no one inside the business owns the workflow design. Fix: ST1512 cohort plus AU0010 governance unit for the digital leadership layer.
Pattern 2, advanced forecasting team, no governance backbone. Sophisticated ML in trading, no firmwide framework for how AI gets deployed elsewhere. Fix: AU0010 across the senior leadership, ST1512 to spread workflow capability.
Pattern 3, store-floor pilot stuck. AI assistants tested in a handful of stores, can’t scale because no one’s built the adoption playbook. Fix: AU0011 for ops directors, ST1512 for area managers, closed-cohort short course for the pilot team.
How TESS is delivering this for retail clients
We run mixed cohorts that combine head-office and field populations. Apprenticeships are sequenced to retail trading peaks (Q4/Q1) so off-the-job time doesn’t collide with peak. AU0009/10/11 units are typically front-loaded against the senior leadership population so the governance frame is in place before the wider ST1512 cohort goes live.
Want a sector-fit briefing for your retail business?
Bring us your AI tool stack, your colleague headcount across head office and stores, and the trading peaks we’ll need to plan around. We’ll lay out the right three-tier stack with the levy maths against it.
Where to read next
Three pieces that round out the retail angle: our Copilot vs Gemini vs Claude UK comparison, our AU0009/10/11 complete guide, and the new Build AI Agents workshop for fast hands-on capability in the e-comm and customer-service teams.
Frequently asked questions.
Do retail apprenticeships work for store managers and area managers?
Yes. ST1512 (AI & Automation Practitioner) is well-fitted to area managers being upskilled into AI workflow design, and to store managers in roles that include cross-channel customer data. The 6-hours-per-week off-the-job time is the binding constraint, most retailers sequence cohorts to avoid trading peaks.
Can we use the levy for our Copilot rollout?
Indirectly. The licence itself isn’t levy-eligible, but the upskilling needed to actually use it productively is. ST1512, AU0010 and AU0011 all draw on the levy and produce the workflow-design and governance capability most Copilot rollouts need.
What’s the right starting point for a retailer that hasn’t commissioned AI training before?
Two parallel cohorts. AU0010 for 6–10 senior leaders (governance and risk backbone, £750 each, ships in 6 weeks). ST1512 for 4–6 trading, merchandising or store-ops generalists (workflow capability, 15 months). The combination produces both the strategic frame and the operational delivery.
Does ST1398 make sense for a retailer?
Yes for retailers with a meaningful in-house data and AI engineering function, typically the larger fashion players, the grocers, and JLP. Best fit: forecasting, personalisation, and supply-chain optimisation teams. Up to £26,000 of levy spend per learner over 24 months.
How do you avoid clashes with trading peaks?
Cohort planning. We sequence off-the-job time around Q4 trading and January sale, with module density backloaded into Q2 and Q3 when retail trading volume is lower. The 6-hour-per-week rule is averaged, not weekly, which gives meaningful flexibility.
Is this just for the head office or does it scale to the store estate?
Both. The head-office AU0009/10/11 stack lands first. ST1512 then runs cohorts in waves through the trading and merchandising population, then through area managers and operations. Most retailers reach meaningful store-floor capability inside 12 months of starting the first cohort.
Related programme: AI & Automation Practitioner Level 4, fully levy-funded with TESS Group.